用人工神经网络预测黑火药燃烧性能

Predicting the Thermodynamic Parameters of Black Powder by Artificial Neural Network

摘要: 利用人工神经网络算法建立了黑火药燃烧热力学参数的定量BP网络模型,通过10组配方的元素组成及其输出参数测试值对模型进行了训练,用另外9组配方的测试结果与相应的预测结果进行了对比研究.结果表明,该方法能较好地对黑火药的燃烧参数进行预测,预测值和试验值误差小于7%,精度较高,可作为功能黑火药配方设计、输出特性参数预测的工具.

Abstract: An artificial neural network(ANN) model about thermodynamic parameters evaluation of black powder was set up.After being trained by a train-set containing 10 compositions,the BP model was used to predict the thermodynamic parameters of black powder,and the predicted values were compared with that of experiments.The results showed that the prediction errors are less than 7%,and the ANN model was capable of making accurate predictions of combustion parameters of black powder.